IT lexicon AI & ML Dataset

Dataset

AI & ML På svenska → Updated: 2026-05-23

The collection of data an ML model trains on — quality and size matter more than architecture.

"Garbage in, garbage out": a model trained on biased data becomes biased. Classic sets: ImageNet (14M images), Common Crawl (for LLMs), MNIST (digits for beginners).

← Back to the lexicon